Hey Nick, yes, have definitely noticed something in Claude. I noticed it leading up to the last big update too. It's just a "vibe" I can't put my finger on. The AI editing process hasn't felt as thorough maybe? I have a couple of prompts I like to use at the very end of an edit as a final catch and they just don't seem to work that well anymore (I don't kid myself that my expertise has improved so vastly in a short period of time, lol). It's a vague disquiet that change is on the way and it feels different.
Thanks for writing to honestly about it - I love seeing the "middle" of the process, so much more interesting than the shiny output where everything is perfection!
This really got me thinking. While I haven’t fine-tuned a writing process to the degree that you have, I do notice what you describe quite regularly. The output across all models changes in a noticeable but un-nameable way. Like you, I put this down to the models shifting. I’m not sure if that is actually the case but as I read its output ‘they must be working on the model’. The ground underneath us keeps shifting.
First, let me say that I’m thoroughly enjoying your series and writing style. Now. a question regarding this column’s subject: Something That Has Not Yet Been Named.
Is this strictly an issue related to Claude Opus 4.6? I use Gemini. Should I be experiencing this condition, as well?
Thanks for a daily well-worth-the-read series. It continues to be a source of inspiration.
Hey Nick, yes, have definitely noticed something in Claude. I noticed it leading up to the last big update too. It's just a "vibe" I can't put my finger on. The AI editing process hasn't felt as thorough maybe? I have a couple of prompts I like to use at the very end of an edit as a final catch and they just don't seem to work that well anymore (I don't kid myself that my expertise has improved so vastly in a short period of time, lol). It's a vague disquiet that change is on the way and it feels different.
Thanks for writing to honestly about it - I love seeing the "middle" of the process, so much more interesting than the shiny output where everything is perfection!
The editing thoroughness drop is something I've noticed too. Like it used to push back on weak spots, and now it just... agrees more easily?
Less friction in the collaboration.
Which sounds good until you realize the friction was the point.
Yes! There is less friction. I ask for things and it does a half-hearted sweep?
When you have it figured out and how to counter-measure, looking forward to reading about it!
Hahaha I'm looking forward to reading that too.
Can someone bring future me back in a hot tub time machine so he can tell present me how to fix this in the present?
😂😂 And, we all have confidence that Future Nick will have this sorted most excellently, which is the most comforting thing in this current daze!
That's the question I don't have data on yet.
My bubble is mostly Claude users, so the signal might be skewed.
You might be the canary. Report back?
This really got me thinking. While I haven’t fine-tuned a writing process to the degree that you have, I do notice what you describe quite regularly. The output across all models changes in a noticeable but un-nameable way. Like you, I put this down to the models shifting. I’m not sure if that is actually the case but as I read its output ‘they must be working on the model’. The ground underneath us keeps shifting.
First, let me say that I’m thoroughly enjoying your series and writing style. Now. a question regarding this column’s subject: Something That Has Not Yet Been Named.
Is this strictly an issue related to Claude Opus 4.6? I use Gemini. Should I be experiencing this condition, as well?
Thanks for a daily well-worth-the-read series. It continues to be a source of inspiration.
-john